The textarea is the source of truth. Tree View is the same parse drawn as nested <details> (see js/tree.js). Objects get names, arrays get 0, 1, 2… Strings and numbers are leaves. I built it because hunting items[7].sku in a 2000-line dump is a waste of a morning.
{
"user": {
"id": 18,
"roles": ["editor", "reviewer"]
},
"active": true
}
You should see user with id and roles, and roles with two indexes. active sits next to user.
Click a node. The side panel fills with a key (or index) and a value you can copy. Nested stuff is JSON in that value field so you can grab a subtree. Expand on the sidebar opens every details at once — useful once, noisy forever.
Turn Tree View on after the JSON parses. Big documents turn into a lot of DOM. If the tab crawls, minify to check it’s valid, then paste a smaller slice. This is not JSONPath and not a diff; two documents belong in Compare.